Update README.md

README.md

@@ -1,33 +1,21 @@
----
-dataset_info:
-  features:
-  - name: id
-    dtype: int64
-  - name: season
-    dtype: int64
-  - name: episode
-    dtype: int64
-  - name: scene
-    dtype: int64
-  - name: line_text
-    dtype: string
-  - name: speaker
-    dtype: string
-  - name: deleted
-    dtype: bool
-  splits:
-  - name: train
-    num_bytes: 5065062
-    num_examples: 47927
-  - name: test
-    num_bytes: 636404
-    num_examples: 5991
-  - name: val
-    num_bytes: 634534
-    num_examples: 5991
-  download_size: 3405920
-  dataset_size: 6336000
----
-# Dataset Card for "the_office_lines"
+## the_office_lines
 
-[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
+<img src="https://a.pinatafarm.com/1351x1232/c8fa71efd1/the-office-handshake.jpg" width="256">
+
+A dataset of lines from the U.S. version of the tv show "The Office". Lines were originally scraped from the website [officequotes.net](https://www.officequotes.net/), are fan-transcribed, and may be of dubious quality.
+
+Contains a train split (47,927 lines), test split (5,991 lines) and validation split (5,991 lines). Contains lines from all 9 seasons, every episode, but may be complete.
+
+
+Lines are annotated with an ID number, season number, episode number, scene number (within the episode), speaker name, and whether or not the text came from a deleted scene. Here is an example:
+
+```
+> dataset["val"][0]
+{'id': 3735,
+ 'season': 2,
+ 'episode': 5,
+ 'scene': 32,
+ 'line_text': 'No, you have the power to undo it.',
+ 'speaker': 'Creed',
+ 'deleted': False}
+```